Effects Of Different Messaging Strategies On Alcohol Use Intentions

Background

Discuss about the Current Directions in Psychological Science.

Development of country is purely dependent on the welfare of the people. There are certain behaviours and phenomenon which might lead to the disorientation of people focus towards having better lives. Alcohol use is among the behaviours and instances which should be controlled to ensure that there is no misuse. Various research has shown that misusing alcohol use leads to reduced productivity of the users. In this cases, it is the responsibility of the government to create approaches to educating the citizens on the dangers of uncontrolled use of alcohol. Due to vested interests from various stakeholders of the business of selling alcohol, it might be more effective to create an independent body which develops and creates strategies on informing the citizens about the dangers of alcohol use. However, it is a personal choice to make the right decision on how and when to use alcohol. Different people have varying versions and reason of why they use alcohol. Some people are sensitive to using alcohol and they tend to limit the use. On the other side, others know the dangers of overusing alcohol but they tend to assume and decide to use carelessly.

Due to the dynamic nature of controlling alcohol use in the society, public health under the government has developed strategies to inform and sometimes instil fear on the users in order to reduce the rate of excessive use. The public health has developed different messaging strategies amongst which most have failed and few have shown significant results in helping the drinkers stay within low-levels of drinking. According to Collymore and McDermott (2016), the fear-based messaging method has been very significant in changing the notions and intentions of using alcohol in the future. In addition, messaging methods which focused on health loss was most effective because the respondents and the population, in general, wish to be in good health. Therefore, any instance which seems to have a negative outcome towards health would be avoided significantly. However, people tend to forget and in most cases, excessive alcohol use is highly directed by psycho-social issues which affect effective decision making on what is good. Research should be well documented and replicated to ensure that people do not rely on false positives(Bonett, 2012). In this study, we focus on finding whether different messaging strategies would influence alcohol use intentions in the future. We also investigate whether gender, age and non-drinking motives would be used in explaining consumption of alcohol in the future.

Methodology

The participants of this study were randomly allocated with a questionnaire to provide their perception of alcohol use. The sample consisted of 500 participants amongst which 33.2% (166) of the sampled participants were male and 66.8% (334) were female. The youngest participant was 18 years and the oldest was 44 year. The average age of the participants was 28.59 with a standard deviation of 3.92.

86 (17.2%) participants were allocated to the loss-fear health groups, 92 (18.4%) on the loss-disgust social group, 71 (14.2%) to the gain health, 86 (17.2%) to the loss-fear social, 87 (17.4%) to the loss-disgust health and 78 (15.6%) to the gain social.

The intentions to reduce levels of alcohol drinking were measured using a 5-level Likert scale; with 1 representing less likely and 5 representing more likely to reduce levels of alcohol consumption management.

This measured the reasons for abstaining from drinking. A 5-point Likert scale was used to measure the reasons for not drinking alcohol from not all important to extremely important(Stritzke & Butt, 2001). The non-motives measure of alcohol use has shown high reliability according to Bekman et al. (2011).

The participants of the study were selected by inviting them through a social media advertising platform. Willing individuals were sent a web link to the questionnaire to fill after they were assigned to an experimental group – the messaging frame. Thereafter, the study was actualized using Qualtrics – which is an online survey system.

	Count	Percent
Male	166	33.2
Female	334	66.8
Total	500	100.0

The sample size of the study was 500 with 166 male and 334 female participants.

Table 2: Message framework conditions frequency table

Condition	Count	Percent
Loss Fear Health	86	17.2
Loss Disgust Social	92	18.4
Gain Health	71	14.2
Loss Fear Social	86	17.2
Loss Disgust Health	87	17.4
Gain Social	78	15.6
Total	500	100.0

According to table 2 above, 29.8% would reduce their level of drinking because of positive reason which includes gaining health and social. The rest (70.2%) would reduce their drinking levels because they fear about their health and social life.

	N	Minimum	Maximum	Mean	Std. Deviation	Skewness	Kurtosis
Intention	500	1	6	2.95	1.176	.487	-.178
Age	500	18	44	28.59	3.920	.177	.313
MAAQ; Nondrinking motives	500	1.00	5.00	3.4755	.96586	-.305	-.704

On average, the drinking intention measure had an average of 2.95 with a standard deviation of 1.176, indicating that most of the participants had greater intention to reduce their levels of drinking alcohol.  Based on the skewness statistic, we can affirm that the intentions variable was slightly skewed to the right because it is positive. The age of the participants had a mean of 28.89 with a standard deviation of 3.92. The skewness statistic is positive indicating that the age distribution was slightly skewed to the right – with more participants being older than the median age.

Results

Table 3: Average of drinking intentions by the message conditions

Condition	Mean	Std. Deviation	N
Loss Fear Health	2.78	.999	86
Loss Disgust Social	2.71	.955	92
Gain Health	2.35	1.057	71
Loss Fear Social	3.03	1.111	86
Loss Disgust Health	4.07	1.108	87
Gain Social	2.62	1.035	78
Total	2.95	1.176	500

On average, positively oriented message conditions had lower drinking intentions as compared to the negative. Individuals who were in the category of loss disgust health had the highest intentions levels of reducing their alcohol intake in the future, followed by the loss-fear social category. The loss fear social had the highest variation of the variation measure and loss disgust social management had the least.  

Table 4: Levene’s test of equality error variances

Dependent Variable:   Intention
F	df1	df2	Sig.
.346	5	494	.885
Tests the null hypothesis that the error variance of the dependent variable is equal across groups.
a. Design: Intercept + Condition

According to the table 4 above, we can conclude that the variation of error terms is not homogenous across all the message conditions because the p-value is greater than the significance level (0.05). Due to the variation in the conditions groups, we can continue to interpret the results of the ANOVA table. 

Table 5: ANOVA test of between-subjects effects

Dependent Variable:   Intention
Source	Type III Sum of Squares	Degree of freedom	Mean Square	F	Sig.
Corrected Model	151.629a	5	30.326	27.793	.000
Intercept	4250.225	1	4250.225	3895.247	.000
Condition	151.629	5	30.326	27.793	.000
Error	539.019	494	1.091
Total	5036.000	500
Corrected Total	690.648	499

According to the table 5 above, the corrected model is statistically significant (p-value <0.001). Therefore, we can conclude that there is a statistically significant difference between the errors of the different message conditions. In additions, the message framework condition variable is a statistically significant predictor of the drinking intentions. The R-squared statistic indicates that 22% of the variation in the drinking intentions can be explained using the message conditions. Further, post-hoc tests need to be performed to justify the differences between the conditions groups.

According to table 6 above, comparing loss-disgust health with all other message conditions were statistically significant with p-values less than 0.001. In additions, the intentions average of gain health and loss-fear social were significantly different at 5% level of significance.

A regression analysis was done to evaluate the effects of age, gender and intention on the non-drinking motives of the participants. Using the Enter method of regression in SPSS, age was entered as the first predictor and gender & intention as the second. The first model with included age as the only predictor had an R-square statistic of 0.00. However, the second model which had age, gender and intention have the predictors had an R-squared statistic of 21.8. This indicates that age is not a significant predictor of the non-drinking motives of the Australians. In addition, the first model was not statistically significant (p-value = 0.932), while the second was significant (p-value = <0.001).

Table 7: Summary of the models

Model	R	R Square	Adjusted R Square	Std. The error of the Estimate
Sig. F Change
1	.004a	.000	-.002	.96682	.932
2	.467b	.218	.213	.85675	.000

Table 8: Summary of coefficients

Model	Unstandardized Coefficients	Standardized Coefficients	t	Sig.	Correlations
B	Std. Error	Beta	Zero-order	Partial	Part
1	(Constant)	3.503	.319		10.992	.000
Age	-.001	.011	-.004	-.086	.932	-.004	-.004	-.004
2	(Constant)	2.565	.324		7.907	.000
Age	-.004	.010	-.015	-.370	.711	-.004	-.017	-.015
Gender	-.069	.081	-.034	-.845	.399	-.021	-.038	-.034
Intention	.383	.033	.467	11.743	.000	.465	.466	.466
a. Dependent Variable: MAAQ; Nondrinking motives

Intention variable is the only significant predictor of non-drinking motives with a p-value

Discussion

The R-squared statistic for the model which includes age, gender and non-drinking motives is 21.9. The overall model is statistically significant with a p-value < 0.001.

Table 9: ANOVA table for the model

Model	Sum of Squares	Degree of freedom	Mean Square	F	Sig.
1	Regression	150.924	3	50.308	46.233	.000b
Residual	539.724	496	1.088
Total	690.648	499
a. Dependent Variable: Intention
b. Predictors: (Constant), Age, MAAQ; Nondrinking motives, Gender

Table 10: Summary of the coefficients

Model	Unstandardized Coefficients	Standardized Coefficients	t	Sig.
B	Std. Error	Beta
1	(Constant)	.606	.418		1.449	.148
MAAQ; Nondrinking motives	.568	.048	.466	11.743	.000
Gender	.091	.099	.036	.917	.360
Age	.008	.012	.025	.636	.525

Non-drinking motives is a statistically significant predictor of intentions of reducing alcohol intake in the future. Age and gender are not significant predictors with p-values greater than the significance level (0.05).

Amongst the aims of this study was to investigate whether message frames would significantly be used to predict alcohol consumption intentions. Using ANOVA between-subjects effects model, it was found that the different message frames would be used to predict intentions of alcohol consumption. The model returned an F-value of 27.793 (p-value = <0.001). Using the message frames, the alcohol consumption intentions can be explained up to 22%. Therefore, we could confirm the research dictates that the message frames define the intentions and decisions of alcohol use. Based on the summary statistics of the averages of alcohol consumption intentions by the message frames, we can observe that the negatively skewed conditions are related to high levels of intentions. This indicates that humans would be more convinced to make a wise decision if they are informed about the negative impacts of a certain action. According to Curtis et al. (2004), human disgust feeling can be used as a method to reduce the risk of a disease. In this manner, using disgust oriented message frames to advertise against excessive use of alcohol, better results will evolve as compared to the use of positively oriented methods.

Further, although the ANOVA model showed that the alcohol intentions are different by the message frames, post-hoc analysis was required to identify the specific groups that are different(Seltman;Howard, 2017). Using Tukey HSD method, we identified several comparisons which were statistically significant. First, the averages of alcohol consumption intentions were statistically significant between loss-fear social and gain health categories. Also, comparing Loss-Fear social and Loss-Disgust health had alcohol consumptions which are significantly different. For all the groups compared with Loss-Disgust health had significantly different average levels of alcohol consumption intentions. Therefore, we would confidently say that the main contributor of the difference in the ANOVA table is the Loss-Disgust health. This indicates that people in the Loss-Disgust health treatment group provided significantly different views of alcohol consumption intentions. The average alcohol intentions of the Loss-Disgust health was 4.07, which is significantly different from the others. These findings are significantly differenced from the findings of Gallagher and Updegraff (2012), who found that gain-framed messages were more effective in promoting health prevention attitudes compared to the loss-framed.

Non-drinking motives were also predicted using a linear regression with age, gender and message frames. In a model with age as the only predictor indicated that it was not a significant predictor. Further, it could not explain any of the variation experienced by the non-drinking motives variable(Gire, 2002). The model had an R-squared statistic of 0.00%, which is a clear indication that 0% variation has explained. Adding gender and message frames in the model improved the model’s R-squared statistic to 21.8%. In addition, the second model was statistically significant with a p-value < 0.001. Message frame was the only significant predictor in the model at 5% significance level. These results coincide with the findings of Stritzke and Butt (2001) who found that drinking motives predicted aspects of alcohol – within which the vice versa is true(Kuntsche, Knibbe, Gmel, & Engels, 2005).

Finally, the alcohol consumption intentions were predicted using linear regression approach.   One model was fitted with three predictors – age, MAAQ; non-drinking motives, and gender. The overall models were statistically significant at 5% level of significance. Gender and age were not statistically significant in the model with p-values greater than 0.05. MAAQ; non-drinking motives variable significantly predicted the alcohol consumption intentions with a p-value < 0.001. The model explained 21.9% variation of the alcohol consumption intentions using gender, age and non-drinking motives. Therefore, the model could be improved by removing gender and age predictors because they are not statistically significant, hence they might reduce the predictive power of the model(Eberly, 2007).

In conclusion, we affirm that base on the study findings, message frames significantly influence intentions of future alcohol consumption. In addition, they can also be used in developing a predictive model to determine the alcohol consumptions intentions of an individual. On the other side, gender and age were found to be insignificant predictors of alcohol consumption intentions. This study was also subject to selection biases. Although the participants were randomly allocated to the treatments, the recruitment was not fair enough to give every individual from the target population an equal chance of participation. Therefore, the study faces threats to external validity because the results might not be representative of the population.

References

Bekman, N. M., Anderson, K. G., Trim, R. S., Metrik, J., Diulio, A. R., Myers, M. G., & Brown, S. A. (2011). Thinking and drinking: Alcohol-related cognitions across stages of adolescent alcohol involvement. Psychology of Addictive Behaviors, 25(3), 415–425. https://doi.org/10.1037/a0023302

Bonett, D. G. (2012). Replication-Extension Studies. Current Directions in Psychological Science, 21(6), 409–412. https://doi.org/10.1177/0963721412459512

Collymore, N. N., & McDermott, M. R. (2016). Evaluating the effects of six alcohol-related message frames on emotions and intentions: The neglected role of disgust. Journal of Health Psychology, 21(9), 1907–1917. https://doi.org/10.1177/1359105314567910

Curtis, V., Aunger, R., & Rabie, T. (2004). Evidence that disgust evolved to protect from risk of disease. Proceedings of the Royal Society B: Biological Sciences, 271(Suppl_4), S131–S133. https://doi.org/10.1098/rsbl.2003.0144

Eberly, L. E. (2007). Multiple linear regression. Methods in Molecular Biology (Clifton, N.J.), 404, 165–187. https://doi.org/10.1007/978-1-59745-530-5_9

Gallagher, K. M., & Updegraff, J. A. (2012). Health message framing effects on attitudes, intentions, and behavior: A meta-analytic review. Annals of Behavioral Medicine, 43(1), 101–116. https://doi.org/10.1007/s12160-011-9308-7

Gire, J. T. (2002). A cross-national study of motives for drinking alcohol. Substance Use and Misuse, 37(2), 215–223. https://doi.org/10.1081/JA-120001978

Kuntsche, E., Knibbe, R., Gmel, G., & Engels, R. (2005). Why do young people drink? A review of drinking motives. Clinical Psychology Review. https://doi.org/10.1016/j.cpr.2005.06.002

Seltman;Howard. (2017). One-Way ANOVA. Experimental Design and Analysis, 171–190. Retrieved from https://www.jmp.com/content/dam/jmp/documents/en/academic/learning-library/04-one-way-anova.pdf

Stritzke, W. G. K., & Butt, J. C. M. (2001). Motives for not drinking alcohol among Australian adolescents: Development and initial validation of a five-factor scale. Addictive Behaviors, 26(5), 633–649. https://doi.org/10.1016/S0306-4603(00)00147-7